In exchange theoretic models assuming self-interest utility maximization the structure of a network biases redistributive outcomes toward the powerful agents. Experiments have shown, however, that powerful agents do not always make use of their opportunities but instead redistribute towards disadvantaged agents. Currently, the rationale of this behavior is unclear. We design an experiment in which we, firstly, systematically vary the relative endowment of the middle player in a 3-line network and test whether the size of a redistributive tax rate favored by this player in the network varies with the endowment and whether this player is able to forge a coalition with one of the other players on that proposed tax rate. Secondly, we test whether the endowment effect interacts with social preferences in order to disentangle positional, social structural, and personal causes of redistributive taxation in experimental groups.
